Date: Thu, 15 Feb 2007 17:50:50 -0600

From: Richard Wright

Subject: Loss of chance

A full answer would take more space and time than is possible or appropriate. I will try to answer, incompletely, with two points.

Proof or disproof of singular instances of causation (or anything else regarding what actually happened on a particular occasion) requires particularistic, concrete evidence that instantiates the relevant causal generalizations or demonstrates their lack of instantiation. Naked statistics, and even ex ante causal probabilities related to general causal capacity (rather than specific causation in the particular instance), merely provide betting odds on what happened or was likely to happen; they provide no information regarding what actually happened in the particular instance (how can they, since one can provide naked statistics and ex ante causal probabilities without any reference to the particular occasion!), which may have been a failure of the extremely likely occurrence to occur (as we know from Barbaro's last horse race) or an occurrence of a very unlikely occurrence. Proof of what actually happened or failed to happen proceeds by providing particularistic evidence that one story (set of causal generalizations) was actually instantiated on the particular occasion, no matter how unlikely its prior probability based on naked statistics of ex ante causal probabilities, and that opposing stories were not instantiated on the particular occasion. This is how lawyers operate in court. The relevant 'probabilities' are unquantified ex post probabilities (judgments) regarding the complete instantiation or lack of instantiation of the various competing stories based solely on particularistic evidence, which at some (undefined) point may produce a subjective belief of varying strength in what actually happened given the sufficiently great ex post probability of one story and the sufficiently low ex post probability of the competing stories. Under the 'preponderance of the evidence' standard, we require only the slightest, minimal belief -- but a belief, nonetheless. It is not mathematics, but psychology.

In uncertain-causation cases, when the defendant may have tortiously caused the plaintiff's injury but it is impossible for actual causation to be proven (or disproven), there may be strong justice arguments for shifting the burden of proof on causation from the actually wronged plaintiff (in Cook or Summers) or possibly wronged plaintiff (in a medical "lost chance" case) to the wrongfully acting defendant who may have caused the injury. If the defendant cannot disprove causation (which requires particularist evidence), it may be fair to impose full liability or, instead, partial liability proportional to the statistical probability of causation. In my view, the choice between these two options -- or no liability -- varies considerably given the different types of uncertain-causation cases, but explaining how and why would require insertion at this point of a long and incomplete paper.

- Richard

-----Original Message-----
From: Benjamin Zipursky
Sent: Thursday, February 15, 2007 2:54 PM
Subject: RE: ODG: Loss of chance

Richard:

I am quite attracted to this view, which essentially relies on two ideas:

(a) various pieces of concrete evidence (E.g., the position of the shooter relative to the victim, the kind of shot used, etc.) to generate a prima facie case of causation for the plaintiff;

(b) the inadequacy of naked statistics to rebut the prima facie case.

I think (a) is vitally important and have said so elsewhere.

My hesitation to jump into this camp rests on my uncertainty over what justifies (b). To say this is not that I disagree with (b), or to say that I think naked statistics are sufficient to prove an element. It is literally to say that it is important to state a theoretical ground for the inadequacy of naked statistics.

One of the more attractive ways to deal with some of the naked statistics puzzles is to say that a plaintiff must do better than this to justify a deviation from the status quo to justify the state cranking up its legal machinery to shift the loss. On the face of it, that does not work when it is the defendant who is trying to use naked statistics in support of the default position.

Another attractive way to justify the naked statistics bar in the plaintiff scenario is to say that naked statistics are really not sufficient to justify the factfinder in believing what the plaintiff asserts. Here again, I am skeptical that this transfers over to the defendant's case. Certainly, in a 100 person version of Summers v. Tice, the fact that it is 99% likely that it was not the defendant Tice seems to me to create a very good reason -- coupled, at least, with Tice's own testimony that it was not him -- to undercut the factfinder's belief that it was Tice. This would apply, in my view, even if the 99 other hunters (Simonsons) could make the same argument.

<<<< Previous Message  ~  Index  ~  Next Message >>>>>